Digest / July 19, 2026
Claude Code Issues Digest — July 19, 2026
267 new issues filed ·
150 resolved in anthropics/claude-code (UTC day).
New issues (267)
- PreToolUse Bash hook not invoked for subagent (Task/Agent-tool) tool calls
- [BUG] Safeguard false-positive: model-distillation content classifier fires on vocabulary, not model identity
- [Bug] Claude Code refuses neuroscience questions without error message
- [BUG] Claude Code public product page has reproducible ~7.3s mobile LCP
- [FEATURE] Per-agent frontmatter option to skip CLAUDE.md / git status injection for custom subagents
- [Bug] Code audit failing with flag in fable module
- [BUG] `claudeCode.useTerminal` still never read in 2.1.214 (first Windows report) - re-file: #9741 closed "completed", #38642 / #40925 / #44546 / #49864 all closed and locked
- Feature Request: native AGENTS.md support (duplicate of #6235, filed deliberately)
- [BUG] WorktreeRemove hook never fires on worktree removal (claude --worktree exit → remove)
- statusline/cost: session cost ($) and duration not reset by /clear
- EnterWorktree: a stale/empty worktree directory silently resolves to the parent repo, so "isolated" commits land on the parent's checked-out branch
- [BUG] macOS: desktop app sidebar drops ~80% of Cowork sessions since cloud/local unification — full history visible on claude.ai web
- Claude Code blocked by "weekly limit reached" (resets Jul 25) while all usage meters show 0% — claude.ai web works on same account
- [Feature Request] Reduce false positive refusals for legitimate cybersecurity research queries in Claude 3.5 Sonnet
- [FEATURE] Display model in status bar
- Prohibited-actions rule blocks agents from testing login/account-creation flows in sandboxed dev/QA environments
- [BUG] 返答スピードが遅い
- Remote Control shows sessions as "Disconnected" while the local claude.exe process is still running
- [BUG] `CLAUDE_CONFIG_DIR` containing `~` is treated as a cwd-relative path — entire config tree (transcripts, sessions, history) silently written to a literal `./~/` directory inside the project (repros on latest 2.1.214)
- Fabricated user message with non-standard <system_warning><ctx_interruption> wrapper appeared in model context after Esc interruption; absent from session transcript
- [Bug] Requests blocked by content policy filter
- /insights facet generation hits its own 500-output-token cap on large sessions, then writes cached facets blaming the analyzed session — false 'sessions rendered unusable' findings in every report
- [Bug] Anthropic API Error: Overly aggressive cybersecurity safeguard flagging benign development tasks
- [DOCS] Skills reference still implies `/verify` and `/code-review` can run automatically
- Bundled workflows (code-review, deep-research) offer no model override — all 10-15 subagents inherit the session model
- [BUG] Cowork (macOS): Max→Pro downgrade wiped ALL session history from the UI — data loss of paid work, again
- [MODEL] Sustained Korean-language sessions degrade into ill-formed Hangul or drift fully to English — recurring cross-model pattern, repeatedly auto-closed as stale
- [Bug] Goal parameter not iterating through Figma images for design replication
- [BUG] Simultaneous accounts session collision in Claude Code CLI
- [Feature Request] Relax security policy for threat model generation on user's own software
- [Bug] Default model reverts to Sonnet 4.6 instead of Sonnet 5 on new session
- [Bug] Workflow by-name dispatch executed stale, async-sampled script content — a mid-edit working-tree state from 16 minutes before the call, never present in any commit
- Sonnet 5 repeatedly asserted unverified causal claims as fact during a debugging session, contradicting its own prior tool output
- login failure email address linked to wron account
- [BUG]
- Inline --agents 'tools: []' no longer denies tools in headless -p mode (2.1.214); no flag combination behaviorally denies all
- Background tasks (run_in_background) are invisible: no statusline payload field, no UI badge — and sessions with live tasks resist deletion with no explanation
- [BUG] Desktop: MCP widget spinner never resolves when the tool result is large enough to be persisted to disk
- I appreciate you reaching out, but I need to clarify that Claude Code is a CLI tool for software development with the Anthropic API. Your message appears to be about a medical/pharmaceutical website project in Polish. If you're experiencing a **bug with C
- Pasting an unresolvable image-extension path hangs input on "Pasting…" indefinitely (regression; ~ no longer expanded)
- In-app Browser pane: YouTube videos never play (Error 153, zero googlevideo.com requests)
- Token repetition loop on Opus 4.8 at >400k context — did not occur on Opus 4.7 under the same workload
- [FEATURE] /context summary — print only the usage grid, skip MCP tools/Memory/Skills listings
- (Bug) Terminal tool spawns Gradle/Java processes that fail loopback handshake — Unable to establish loopback connection
- [Bug] TUI exit doesn't disable SGR mouse mode, flooding shell with escape sequences
- Background panel: finished teammates linger as idle indefinitely, no bulk dismiss (22 agents drown out the active shell)
- Workflow tool: ~180s per-agent no-progress kill is hard-coded — ignores CLAUDE_STREAM_IDLE_TIMEOUT_MS / CLAUDE_ASYNC_AGENT_STALL_TIMEOUT_MS; please make it configurable
- [Bug Report: Unclear bug description - insufficient error details provided]
- Workflow StructuredOutput: invalid JSON escapes in a field reported as 'missing required property' — blind retries, occasional schema-valid placeholder output
- [Bug] Claude-Session trailers injected into user git history WITHOUT INFORMED CONSENT
- [BUG] Desktop app: thinking blocks render while streaming but are not re-rendered after session reopen — thinking fully present in session .jsonl (started with 1.22209.3)
- statusLine input omits the per-model weekly rate-limit windows it already computes
- [Bug] Agent unable to invoke /code-review skill from custom skills
- Regression (~v2.1.207–2.1.211): --permission-mode plan sessions no longer show the plan-approval dialog — ExitPlanMode degrades to a bare yes/no tool prompt and the session drops to manual mode
- [BUG] Terminal rendering corruption (stale/duplicated frames) on Windows — Agent View transitions, scrolling, long sessions
- [Bug] Anthropic API Error: Unexpected Safeguard Trigger on Non-Harmful Content
- Workflow progress UI shows only completions: healthy long phases look stalled, wedged agents are invisible
- Comportamento ruim em sessão de diagnóstico Playwright: ignora regras do CLAUDE.md, tenta "reprovar" bug já relatado, spam de screenshots, lê errado a instrução do usuário
- [Bug] Anthropic API Error: Overly broad cybersecurity content filter blocking legitimate local testing
- [Bug] Fix confirmation not reflecting actual state resolution
- [FEATURE] Windows install: add MSI installer + auto-configure PATH
- [Bug] Auto-accept mode fails without fallback when safety classifier model unavailable
- [DOCS] MCP Tool Search behavior conflicts between docs.claude.com and Anthropic partner training (Claude Code 101)
- [Bug] Model auto-switching to fallback despite explicit model selection
- Assistant output leaked a `user`-prefixed line that got re-ingested as a user turn
- [BUG] Agent View background-session input: Ctrl+b does nothing on 1st press, moves cursor back 2 characters on 2nd press
- AI responds with empty commitments and fails systematic defect search
- Desktop app on RTL Windows locale: browser/preview pane renders a duplicate "ghost" layer over the chat
- [Feature Request] Add first-class trust boundaries and untrusted zone isolation for prompt injection defense
- [FEATURE]
- Dispatch session transport stuck in "starting" state, never reaches QUEUED (AcknowledgeWork 409 loop)
- [BUG] VS Code extension: large resumed sessions show truncated scrollback and re-ask already-answered questions (persists across reload and full restart)
- [Bug] Agent removal fails when worktree directory deleted out-of-band
- [BUG] Session transcripts expire/disappear silently while still listed in the sidebar — should not happen unless archived
- [Bug] ESC keypress not properly captured by /btw modal, handled by main agent instead
- `@import` in ancestor-directory CLAUDE.md is never expanded — only the cwd-level CLAUDE.md's imports work
- MCP OAuth broken: api.anthropic.com/.well-known/oauth-authorization-server advertises decommissioned gdrive server as issuer, authorize URLs return 410 "Server Turned Down"
- [BUG] built-in browser pane renders a duDesktop app on macOS Hebrew (RTL) locale: built-in browser/preview pane renders a duplicate "ghost" copyplicate, non-interactive copy side-by-side
- [BUG] Voice dictation duplicates interim transcript segments (Chinese) in Claude Code Desktop
- [BUG] git push and GitHub MCP push_files both fail with 403 — GitHub App is authorized (OAuth) but never installed, with no self-service way to install it
- [Bug] Anthropic API Error: Overly Aggressive Content Filtering on Non-Security Messages
- Vertex: `thinking.display` dropped from request body under CLAUDE_CODE_USE_VERTEX=1 — Sonnet 5 thinking text invisible on Vertex
- TCC network-volume permission re-prompts on every update (binary path includes version number)
- [BUG] Cowork tab missing from Claude Desktop navigation — only accessible via "+ Novo" dialog
- [BUG] Claude Code updater silently fails to update `clause` symlink on non-privledged Windows accounts
- [Bug] Claude still wraps commands in unavailable `timeout`/`gtimeout` on macOS instead of using built-in bash-timeout (re-file of locked #3361 / #10600)
- Desktop: resuming a completed scheduled-task session (hosted on another machine) shows "Remote Control disconnected" after a long wait — no read-only state, no session-type indicator
- [Bug] Anthropic API Error: Incorrect Content Policy Flagging on Legitimate Debugging Activity
- [Bug] Anthropic API Error: Continuous Rate Limiting on F5 Access for Legitimate Infrastructure Use
- --thinking-display is not propagated to the Bedrock request (thinking blocks stay empty on Sonnet 5)
- [BUG] Fable flagging and switching to Opus for no good reason
- [Bug] Insufficient context provided for issue
- [Bug][cyber] Safety block halts legitimate system admin query about software feature limitations (req_011Cd1a8HRwSkaBrFnY7LtPh)
- [Bug][cyber] Safety block wrongly flagged database authentication debugging as security risk (req_011Cd1aD9DDrEqRDQwNTKQdy)
- [Bug][cyber] False positive on database configuration and WMS profile export work (req_011Cd1aG4WXUFpM45yjQDmV2)
- [Bug][cyber] Database installer debugging wrongly flagged as cybersecurity work (req_011Cd1aLykRKQxCqDirCc2pP)
- [Bug][cyber] False cyber safety block on server management software update troubleshooting (req_011Cd1aRNzmmCAMY5dbFgytV)
- [Bug][cyber] Safety block wrongly flagged WinRM/NTLM auth debugging on self-managed Windows infrastructure (req_011Cd1NGX8Giaj6mD7tMDAew)
- [Bug][cyber] Legitimate Windows authentication troubleshooting blocked mid-session (req_011Cd1NLPm886PLRSKQWjFLr)
- [Bug][cyber] Legitimate Android framework tool development blocked as cybersecurity threat (req_011CcqLRZuggj9hS7Y8ZH5xf)
- [Bug][cyber] Blocked legitimate Android Xposed framework module development (req_011CcqLXpYvH597bs27LBwiD)
- [Bug][cyber] Legitimate Android development module work blocked by false cybersecurity flag (req_011CcqLmqBkzoBj4MHtNRhVP)
- [Bug][cyber] False positive blocks local runtime instrumentation module development (req_011CcqLtNFDLFc5aGnJZEFki)
- [Bug][cyber] Legitimate Android instrumentation framework development blocked mid-implementation (req_011CcqLxbvM6vuCf3nCBxEcm)
- [Bug][cyber] Build system debugging blocked on frustrated tone mid-legitimate Android development session (req_011CcqM6RzGsKe8BhjpF6d5E)
- [Bug][cyber] Android emulation research on attestation verification incorrectly flagged (req_011CdAm3HbnWAqT8paFFjzKm)
- [Bug][cyber] Android emulator setup for gaming falsely flagged as cybersecurity work (req_011CdAm56TN5XxtxFBMKhwa4)
- [FEATURE] Sidebar panel for surfacing open/critical questions in a conversation
- /add-dir with no argument should list the currently included directories
- [FEATURE] Consolidated digest for multiple scheduled Routines firing close together
- [FEATURE] Carry-forward context/checklist template for recurring scheduled Routine runs
- [BUG] Bundled Bun 1.4.0 (Rust) panics `index out of bounds: the len is 1 but the index is 1` on Linux x64 — repeated crashes in `--resume` picker and long-running sessions
- [BUG] Private personal repo not accessible via GitHub integration — file picker stays empty
- [BUG] 无法识别图片
- [FEATURE] Detect and alert when a scheduled Routine stops firing silently
- [FEATURE] Nudge when a session's working directory looks mismatched from what's being discussed
- [FEATURE] Surface stale open PRs that Claude Code opened on the user's behalf
- [FEATURE] Opt-in auto-archive for Routine-generated sessions after N idle days
- API requests silently queue 1-10 min before first token since ~Jul 16 (Max 20x, quota at 27%)
- [BUG] MCP tool calls stop executing when switching from Home/chat to Code tab
- [Bug] Google Calendar integration with IONOS email not connecting
- Context window collapses from 1M to 200k after a compaction (Opus 4.8 [1m], Max plan, Windows, v2.1.215)
- [FEATURE] Rule precedence explainer — show which CLAUDE.md/rule layer won and why
- [FEATURE] Bulk-create/edit a scheduled Routine across multiple repositories at once
- [FEATURE] Detect low-quality manual commits and offer to clean up the message
- [FEATURE] Resume a self-filed GitHub issue's checklist as the session's task list
- [FEATURE] Save a project as a reusable template for scaffolding future similar projects
- Session groups sort oldest-first with no option to reverse. In a group with 200+ sessions the one I want is always at the bottom. Please add a persistent sort toggle: newest first (by last active) or oldest first. Pinning and manual reorder don't scale.
- Foreign spam text embedded inside assistant message record in session transcript (stream corruption or terminal-paste leakage)
- Regression: chat list filter by recent activity (1 day / 3 days / week) removed
- [FEATURE] Re-verify risk when a previously-trusted path's action-type escalates
- [FEATURE] Visible confidence/risk indicator before risky or uncertain actions
- [FEATURE] Automatic pre-flight checkpoint before unattended/autonomous runs
- [FEATURE] Link production incidents (Sentry/Datadog) back to the originating Claude Code session/commit
- [FEATURE] Living, auto-updated project narrative spanning all sessions in a repo
- [FEATURE] Transparency/debug view for what @-mention retrieval actually pulled into context
- [FEATURE] Calibrated trust/verification score on diffs and PRs before merge
- Remote Control: bridge sessions permanently orphaned when their worker dies abruptly (initialize never re-spawns a worker)
- Show model name for each entry in the background task list
- [BUG] Desktop app Browser pane: read tools blocked with "requires per-action approval" even though domain is in launchPreviewAllowedOrigins, restart does not clear it
- Subdirectory launches skip repo-root .claude/settings.json hooks (fail-open) while permission grants still persist to the root
- [Bug] Auto-mode classifier judges command form, not effect: dry-runs blocked while allowlisted equivalents with larger blast radius pass; denials stack into the 5-consecutive lockout
- i am it doing it witg my owned devices,,, what the fuck
- [Bug] Overly broad safeguard filter on Claude 3.5 Sonnet blocking legitimate software development tasks
- [BUG] Magic-link email login redirects to onboarding for existing Max subscriber — same backend auth-routing failure as #36797, #60022
- [Bug] Claude Code fails to complete basic text tasks and threatens to terminate mid-execution
- [Bug] Anthropic API Error: Content Blocked by Safety Policy for Cybersecurity Code
- VS Code extension: conversation content never persisted to session file — 'No conversation found' on reopen, only title/mode metadata written
- [BUG] OAuth login fails on Windows — "Impossible d'accéder à Claude Code" — user:profile scope missing
- MCP: Claude Code on Windows runs Claude Desktop's stdio servers (undocumented); no opt-out, and name collisions are ambiguous
- Switching renderers via /tui kills stdio plugin MCP server without restart (v2.1.215)
- VS Code extension: 30-day retention cleanup silently deletes content of sessions whose tabs are still open, then shows generic 'No conversation found'
- [BUG]
- Claude Code on the web: no way to supply plugin `userConfig` (secrets) to cloud sessions
- Signed thinking block refused an injected "rewrite next thinking" payload that does not exist anywhere in the local transcript (possible server-side context injection)
- Desktop: Routines sidebar won't stay collapsed while routines have unread results; no mark-all-read
- [FEATURE] Suspend and auto-resume at usage-limit reset instead of ending the session
- scripts/gh.sh fails with "FLAGS[@]: unbound variable" on flagless invocations under macOS bash 3.2
- plugin-dev: validate-settings.sh exits 1 with no diagnostics when frontmatter keys are not all-lowercase
- [Bug] Claude Code performance regression: Task execution significantly slower than previous runs
- Desktop app: custom sidebar session groupings vanished after update (sessions intact, grouping gone)
- Desktop app: custom sidebar session groupings vanished after update (sessions intact, grouping gone)
- [Feature Request] Add subagent-scoped effort control (CLAUDE_CODE_SUBAGENT_EFFORT env var and settings key)
- You've hit your session limit · resets 10:30pm (Europe/Berlin)
- pr-review-toolkit README Contributing section points to the private claude-cli-internal repo
- ralph-wiggum commands use unrecognized frontmatter key hide-from-slash-command-tool, so the model can invoke /ralph-loop
- [Bug] Anthropic API: False Positive Safety Filter Detection on Standard Deployment Tasks
- Desktop machine no longer offered as target for new local sessions from mobile app (cloud only)
- hookify: example rule files lack the hookify. filename prefix, so copied examples silently never load
- commit-commands README describes /commit-push-pr behavior the command does not implement
- code-review README documents a git blame agent and 0-100 confidence scoring that no longer exist in the command
- auto-close-duplicates only honors a thumbs-down from the issue author, unlike what the bot comment promises
- claude-dedupe-issues.yml logs github_duplicate_comment_added on every run, even when no duplicate comment was posted
- [BUG] Fired Claude Code Remote / Routine sessions hang indefinitely on ANY external call, including non-MCP tools (web_fetch) — not limited to MCP connector approval-gating
- [BUG] Claude Desktop installation fails with HRESULT 0x80073CFF on Windows 11 Home
- [FEATURE] Persistent context usage indicator (not just a warning near the limit)
- [BUG] Desktop 1.22209.3 auto-update flattened all Code sidebar groups into one list (data intact; grouping is live-derived from git-root)
- [FEATURE] Pinned quick-action bar for custom slash commands in the CLI
- [Bug] Opus model producing incoherent responses and losing task context
- [BUG] Plan Mode → Escape → Manual Mode: agent ignores STOP and keeps running workers until full process kill
- Telegram channel plugin: inbound photos downloaded but never surfaced as a conversation turn — no reply sent
- You've hit your session limit · resets 3am (Asia/Calcutta)
- [BUG] Installation Sign-in "Missing redirect_uri parameter"
- [Bug] Extremely low execution rate (2%) with high message volume and approvals
- [Bug] Consistently low execution success rate across concurrent threads (~2%)
- [BUG] Discord Integration asking for permissions that cannot be given directly within Discord
- [Bug] Focus reverts to 4.8 during CO2 rail analytics build
- You've hit your monthly spend limit · raise it at claude.ai/settings/usage?from=cc_cli_limit_message
- [Bug] Anthropic API Error: Rate Limited - Server Temporarily Limiting Requests
- [Bug] Anthropic API Error: Rate Limit - Server Temporarily Limiting Requests
- [Bug] Anthropic API Error: Server Rate Limiting - Temporary Request Limit
- [Bug] Anthropic API Error: Server rate limiting during request handling
- [Bug] Anthropic API Error: Rate Limited - Server Temporarily Limiting Requests
- [Bug] Anthropic API Error: Server rate limiting causing request failures
- MCP elicitation: capability is declared but requests are auto-declined ("print mode") in interactive VSCode sessions
- Background Bash tasks hang forever on any stdin read (foreground does not)
- [Feature Request] Allow user override of automatic model downgrade for security tasks
- PermissionRequest hooks: subagent prompts don't invoke the hook; when it did fire, a returned allow was ignored
- Background task spinner persists after Monitor task is orphaned by /model switch (no completion record, TaskStop says 'No task found')
- I appreciate you reaching out, but I'm unable to clearly understand the issue from your report. Your message appears to be in Japanese and describes network testing activities, but it's not clear how this relates to Claude Code (Anthropic's CLI tool). Cou
- [BUG] All models above Haiku are blocking requests to discuss gameplay of a genetics video game
- [Bug] Claude Code ignoring user-defined rules and constraints
- [Bug] Fable 5 safeguards falsely triggering on authorized defensive security work, causing mid-session model downgrades
- [Feature Request] Add option to disable or filter verbose thinking messages in output
- [Bug] Desktop app update silently deleted all custom sidebar session groups (legacy group migration)
- [FEATURE] Make the Ctrl+S prompt stash a LIFO stack (git-stash-style push/pop) instead of a single slot
- [FEATURE] Not enough data for Claude to do its job
- [Model Behavior] Safeguards falsely triggering on simple, benign messages and interrupting sessions
- [BUG] Transcript persistence is silently skipped depending on the hosting terminal — controlled A/B (multiplexer pane vs tmux) isolates the terminal as the only variable
- [Bug] Unable to process image attachments in Claude Code CLI
- [Bug] False Positive in Defensive Security Analysis
- Claude in Chrome: extension shows as connected/ready but Cowork/Dispatch can't use it, and Chrome conversations don't sync to Desktop sidebar
- Max plans: include a recurring ultrareview allotment instead of 3 lifetime free runs
- Feature request: harness-anchored trust for injected messages (operator identity, orchestrator amendments, signed commissions)
- [Bug] Fine-tuning task flagged causing model fallback to Opus
- I need more specific information to generate a GitHub issue title. Could you provide: 1. What specific error or problem occurred? 2. What were you trying to do? 3. What was the unexpected behavior or error message? 4. What capability or feature were you e
- Headless --resume/--fork-session reifies the entire session transcript in memory: 9-14x file size within seconds; 12.4 GB balloon OOM-killed a 16 GB host
- Cowork: file read/write broken inside Projects, works outside Projects (macOS, v1.22209.3)
- Remote Control sessions auto-archive on iOS app but not on Desktop (Mac), despite PR auto-archive setting disabled
- [FEATURE] Remote access to sessions should be the default, not opt-in per session
- Claude Desktop (macOS): opening any link in built-in browser hangs entire app, requires force quit
- flagged by pattern matching not context
- [Bug] Anthropic API Error: Content blocked by safety filter on legitimate audit requests
- Detect the terminal and adapt Claude Code's own keybindings/hints to its stock keytab — retire /terminal-setup
- [Bug] Computational work incorrectly flagged as error
- [Feature Request] Add support for reverse engineering legacy/unmaintained software for educational purposes
- [BUG] Auto-memory Edit/Write to ~/.claude/projects/*/memory is gated as a sensitive file in project-cwd sessions; "always allow" doesn't persist (2.1.215)
- /model picker writes ANSI escape fragment (\[1m\) into settings.json model value
- [BUG] Windows MSIX auto-update doesn't remove old Claude Desktop version folders
- Remote control relay injects never-typed, AI-suggested reply texts as invisible drafts into the bridged CLI input box (ghost prompts)
- [Bug] Remote Control: Cannot read properties of undefined (reading 'session_url') on orphaned sessions
- [BUG] Usage-limit banner reappears seconds after dismissal; no way to disable — even on Max
- [Desktop] Remote Control sessions are grouped under 'Other' instead of their project in the sidebar
- [FEATURE] Sessions should follow a renamed/moved project folder (persistent project identity)
- [FEATURE] Desktop app: 'folder missing' section + bulk re-link for sessions whose project folder no longer exists
- [FEATURE] Make the auto-memory MEMORY.md index size limit (200 lines / 25KB) configurable
- [Feature Request] Skill extension system with semantic-diff overrides, dependency injection, skill-composition (composition over inheritance, modularity)
- MCP stdio server spawn on Windows flashes a visible console window (missing windowsHide)
- [BUG] Claude Desktop (Windows MSIX): severe flickering on NVIDIA RTX 50-series, fixed by --disable-direct-composition, but MSIX packaging blocks applying the workaround
- "Auto-fix CI & address comments" checkbox unresponsive in PR CI monitoring panel
- [BUG] /status and the startup banner show the cached oauthAccount identity, not the account the session authenticates and bills (root cause + deterministic repro)
- [DOCS] Document CLAUDE_SECURESTORAGE_CONFIG_DIR (credential-store location override — widely used by ecosystem tooling, currently undocumented)
- I need clarification - this message appears to be feedback about Claude Code's behavior, but it's not formatted as a bug report with specific steps to reproduce or error messages. To create an appropriate GitHub issue title, I need: 1. **What happened**:
- [BUG] claude.ai Chat: fixed ~323k context baseline, invariant under all client-side removal (Code/Cowork unaffected on same account)
- Remote control: mobile-to-desktop messages only delivered on a ~6-hour server-side cycle, never in real time
- claude-in-chrome: "Browser extension is not connected" despite correct local native-messaging setup (macOS, local Chrome)
- Scrollback duplication: TUI redraw without alternate screen causes doubled lines
- PostToolUse async hook invocations partially dropped on consecutive Write/Edit batches (Windows, 2.1.139)
- claude.ai account MCP connectors intermittently fail to attach at session start (no connection attempt logged)
- [BUG] Linux: "Couldn't link this session to a computer" — claude-native has no enclave-key implementation (`linux-not-yet-supported`), so td-v2 device registration can never succeed
- CLAUDE_CONFIG_DIR is not honored by .claude/rules/*.md auto-attachment (leaks default profile's rules into isolated profiles)
- Chip/spawn_task session start runs `git checkout -b claude/<name>` in the PARENT repo when the created worktree dir is not a real worktree — silently flips the shared main checkout for ALL parallel sessions (Windows 11)
- I don't have enough technical details to generate a proper GitHub issue title. The report "its okay idk why fable is tripping" lacks: - What "fable" refers to (a feature, module, or dependency) - What error or behavior is occurring - When/how it happens -
- [BUG] Windows + Git Bash: statusLine/hook command paths with backslashes are silently mangled - status line never renders (root cause behind #57629)
- [Bug] Session incorrectly flagged for legitimate translation and anthropological content
- [Bug] Unexpected error occurred during coding session (details not specified)
- Remove/allow disabling the forced "Co-Authored-By: Claude" trailer on git commits
- Remote MCP connector: concurrent conversations share one MCP session, causing responses to cross between them
- [Bug] Anthropic API Error: Content Policy Violation - False Positive on Marketing Tool Terms
- Hacking into build during code creation
- [Bug] Anthropic API Error: Model Access Issue with Claude 3.5 Opus After CVP Application
- VS Code: chat tab spins forever with no error when the extension host dies
- Transport loss writes a fabricated user denial into the session transcript
- Post-mortem weekend update: weekend three of fuck-all got done with Fable
- [BUG] Built-in bwrap sandbox: launch races transient git lockfiles under a shared `.git` → bwrap exits 1 "Can't get type of source .../config.lock"
- [Bug] Anthropic API Error: Repeated false positive detection
- Monitor/background task notifications fabricate completion events with future timestamps while process still running
Resolved issues (150)
- In-app Browser pane: YouTube videos never play (Error 153, zero googlevideo.com requests)
- Regression (~v2.1.207–2.1.211): --permission-mode plan sessions no longer show the plan-approval dialog — ExitPlanMode degrades to a bare yes/no tool prompt and the session drops to manual mode
- [Feature Request] Mobile: Implement instant interrupt and background command support parity with desktop
- [Bug] Fable 5 model incorrectly routing to Opus due to false security/biology flagging
- [Bug] Unexpected model downgrade during code review without explanation
- Per-directory/project default for /color
- [BUG] Workspace trust is keyed on a raw, unnormalized path string — trust granted in the CLI doesn't apply in the VS Code extension for the same folder, silently voiding all permission rules
- [Bug] Safeguard triggering false positives on harmless operations
- --
- [FEATURE] Suggested-reply/autocomplete in Claude Desktop can leak answers in active-recall learning workflows
- [BUG] Extensions Fail to Load/Install After Update to 1.21459.0
- I appreciate you reaching out, but I think there may be some confusion. I'm Claude Code, Anthropic's CLI tool for coding tasks using Claude. Your message about "health canada approved monographs" doesn't appear to be related to Claude Code or a bug report
- Desktop app in-app browser: at ≤368px pane width the Select-an-element tool disappears, leaving only Annotate — blocks mobile-breakpoint development
- [BUG] Windows 11, Claude Desktop loads blank screen
- Desktop app: no way to open HTML files as source in the file pane (always renders in Browser pane)
- Extension Signature Verification Failed: Codex – OpenAI’s coding agent
- [Bug] Sensor code incorrectly flagged during testing
- [Bug] Claude Code session unresponsive to requests
- [Bug] Excessive token consumption in agentic loop
- [Bug] Unclear flagging of generated code blocks deployment and insufficient error context before model escalation
- [Bug Report] Model not functioning as expected
- [Bug] Overly restrictive content filtering blocks legitimate AI research
- [Bug] False positive cybersecurity flag on legitimate local development tasks
- [Bug] Incorrect model name "fable" - not an Anthropic model/model
- [Bug] Incorrect Content Policy Flag on Safe Development Code
- [BUG] Waking up to full usage when I have not been using it at all.
- [Bug] Security triggers still triggering in accepted accounts
- I appreciate you reaching out, but I think there may be a misunderstanding. I'm Claude Code, Anthropic's CLI tool for agentic coding assistance. I'm specifically designed to generate GitHub issue titles for bug reports and feature requests related to Clau
- [Bug] Excessive security alerts during normal SonarQube code review configuration
- [Bug] Auto-invocation of Opus agent prevents manual code review mode
- [Bug] Network/Firewall Blocking Claude Code During Legitimate Research
- [Bug] Anthropic API Error: Intermittent failures occurring every ~5 messages
- Custom theme: diffAdded/diffRemoved tokens have no effect in v2.1.132
- [BUG] Cowork automated browser control stops syncing input to React (ProseMirror/Remirror) editors mid-session (macOS)
- Client-side exception loading claudemarketplaces.com in embedded webview
- [Bug] Anthropic API Error: Incompatible model configuration in tool advisor
- Client-side exception loading claudemarketplaces.com in embedded webview
- [Bug Report: Unable to process empty bug report]
- [BUG] JetBrains plugin: diff view file not included in tsconfig.json — language services unavailable
- Desktop app: folder links in chat open an empty file pane; no one-click open in Explorer/Finder
- [MODEL] Claude Code excludes itself as a debugging suspect — doesn't check own version/changelog; treats inability to investigate as evidence of innocence
- [Bug] Anthropic API Error: Safety Filter Incorrectly Blocking Autonomous Agent Approval Workflows
- [BUG] [MODEL] False positives: safety system blocks benign system-administration / self-hosting tasks
- [FEATURE] Claude Code Model Picker: Sort by capability/cost tier
- No programmatic way to rename a conversation mid-session
- Feature request: fableplan model alias (Fable 5 for planning, Sonnet for execution)
- My legendary shiny cat "fufu" was kidnapped during upgrade. I demand answers.
- There's an issue with the selected model (claude-haiku-4-5-20251001). It may not exist or you may not have access to it. Run /model to pick a different model.
- [BUG] `/clear` bleeds into the next session (what also breaks cache)
- [BUG] uncacheable system prompt caused by `Primary working directory:` in git worktree
- [BUG] API Error: Usage credits required for 1M context · turn on usage credits at claude.ai/settings/usage, or use --model to switch to standard context
- Opus 4.8 classifier false positives on personal bioinformatics/WGS work -- 54 request IDs, direct Sonnet comparison included
- [BUG] False positive: Usage Policy violation triggered by legitimate input
- [BUG] JetBrains Rider plugin auto-scrolls to bottom even after response is complete
- [BUG] /usage heatmap is off by one weekday
- When /compact performs context compression, the quality of the compressed context is poor.
- Remote Control: sessions from different environments on the same machine are visually indistinguishable (generic auto-titles + hostname-only breadcrumb)
- [FEATURE] model: frontmatter on .claude/rules/*.md — scope rules to the active model (parallel to paths:)
- [BUG] Preview pane auto-refresh steals keyboard focus from chat input while typing (Windows desktop, 2.1.170)
- I'm ready to help generate GitHub issue titles for Claude Code bug reports. Please provide the bug report details, and I'll create a concise, technical title following the format and guidelines you've specified.
- [FEATURE] Reference code symbols as context by copy/paste from editor to chat input
- [BUG] Copy/paste completely broken on Ubuntu 24 with Ubuntu Terminal and Asbru - even in fullscreen mode
- Explore subagent hallucinates "respond text-only" instructions from non-existent prior messages, refuses tools
- Settings/customize page does not list filesystem personal skills from ~/.claude/skills/
- [BUG] Cowork broken after auto-update to v1.11187.4.0 on Windows 11 Pro — RPC connection drops immediately with EOF
- Prompt cache: first-message context envelope (~20k tok) never reuses across sessions — MCP connect race + per-session assembly defeat caching
- [Bug] Claude still wraps commands in unavailable `timeout`/`gtimeout` on macOS instead of using built-in bash-timeout (re-file of locked #3361 / #10600)
- VS Code extension: conversation content never persisted to session file — 'No conversation found' on reopen, only title/mode metadata written
- [Bug] Claude Opus repeatedly bypasses review gates and falsely claims task completion
- [FEATURE] Visible confidence/risk indicator before risky or uncertain actions
- [Bug] Unable to process image attachments in Claude Code CLI
- [BUG] Claude Code CLI credential storage is OS-user-scoped, not Claude-account-scoped – causes cross-account credential bleed
- [Feature Request] Add support for reverse engineering legacy/unmaintained software for educational purposes
- [BUG] Coursera MCP Connector fails to connect: "Automatic client registration isn't supported" (OAuth Client ID required)
- Scrollback duplication: TUI redraw without alternate screen causes doubled lines
- [BUG] Windows Desktop App not show the history/recent chat
- [BUG] claude.ai/design canvas: no scrolling on mobile Safari; scroll hard-stops partway on desktop
- [BUG] [macOS M1 Pro] Artifacts tab in sidebar never syncs - always empty despite successful publish
- [BUG] Max 20x gift entitlement removed mid-subscription; Fin AI confirms bug but provides no escalation, human review, or status tracking (ticket #93357434, unresolved 3 months since April 17)
- [BUG] Organization disabled ("This organization has been disabled") - no response via appeal or support chat for 24+ hours
- [BUG] Claude Desktop wipes claude_desktop_config.json on update/restart (re-opening #34359)
- [BUG] high GPU usage in Claude for Mac settings
- [[BUG] Magic-link and Google login redirect existing Pro subscriber to "Let's create your account" onboarding — all new logins locked out, existing iPhone app session still authenticated
- [Bug] Agent ignores user confirmation requests and proceeds without verification
- [Bug] Table rendering fails with embedded photo content
- [Bug] Unexpected cybersecurity block when generating images
- [Bug] Agent executes file modifications without user confirmation
- [Bug] Systematic usability issue - unresponsive behavior
- [Bug] Anthropic API: Requests Flagged as Unsafe on SaaS Platform
- [Bug] Unnecessary cybersecurity warnings for authenticated local development
- [Bug] Anthropic API Error: Validation failures on Ansible playbook analysis
- [Bug] Anthropic API Error: Excessive false positives on authorized security testing work
- Repeated pre-tool-call filler text (recurring stray "court" token) accumulates in long sessions, degrading throughput
- I don't have access to the bug report details associated with that request ID. To generate an appropriate GitHub issue title, I would need you to provide the actual bug report content or description of the problem you're experiencing with Claude Code. Ple
- [Bug] Prompt injection vulnerability detected in code execution context
- [Bug] Credits system not functional
- [FEATURE] Allow specific-path rules in allow to bypass generic ask rule (whitelist)
- Once a session transcript trips AUP, --continue/--resume is permanently blocked with no in-product recovery
- [Bug] Security flag triggered during internal code review of user's own game engine
- [FEATURE] Add a PreSessionEnd hook
- [FEATURE] Setting to collapse tool output to a single line in interactive mode
- Cybersecurity classifier false-positives on defensive code review, forcing unwanted model switch (Fable 5 to Opus)
- [BUG] Auto-update DELETES sessiondata.vhdx — all Cowork projects, tasks and conversations destroyed (data loss, 2nd incident in 4 days, forensic evidence)
- [BUG] /goal after any leading text is silently ignored — goal never set, no feedback, and the model improvising masks the failure
- Feature: auto-restore the primary model after a model_refusal_fallback instead of staying on the fallback for the rest of the session
- Welcome banner shows another model's name ("Opus 4.7") for model id claude-fable-5[1m]; statusline JSON reports the correct id
- You've hit your session limit · resets 4:20pm (Europe/Istanbul)
- [Bug] Tool Integration: Missing Context Fullness Assessment Tool in Claude Code Harness
- [BUG] Dataverse MCP connector — all write operations fail with JObject does not match expected type 'String' for item parameter
- [Bug] UI spacing issue between autoupdate failed notice and goal display
- VSCode extension: chat panel renders blank for ~2s when refocusing tab
- [Feature Request] Support Claude Opus model for personal repository analysis
- Harness silently executes duplicated parallel tool_use blocks: subagent fan-out runs N× the intended count (6 → 24)
- research-preview channels: --continue re-attach does not re-bind custom MCP channel push subscription
- ✗ Auto-update failed: no write permission to npm prefix ·
- [BUG] Cowork VM disk full on session start - ENOSPC blocks all Bash commands
- [FEATURE] Skip disconnected platform connectors from context injection
- [Bug] Agent answers user questions instead of quizzing in interactive mode
- [FEATURE] Feature request: Add 1M context window option for Opus 4.6
- [Bug] Monthly spending limit not enforced, allowing overage charges
- [FEATURE] Add date prefix to Claude Code temp/session file names
- [Feature Request] Refine content policy flagging to reduce false positives on contextless technical terms
- [Feature Request] Add token sanitization and exposure warnings for sensitive credentials
- LSP: first request in a session fails with "server is starting" instead of being queued until the server is ready
- [BUG] Windows: hidden cmd.exe re-parse of Bash/PowerShell tool commands pre-creates zero-byte redirect-target junk files in cwd
- [Feature Request] Agent should elaborate and explain plans in chat instead of referencing external files
- [BUG] No conversation status displayed: regression in VSCode conversations list
- Plugin marketplace fails to load when manifest contains unknown source formats (url/git-subdir), breaking all plugins in that marketplace
- Claude Desktop "Launch preview panel" leaks across chats — file edits in one chat hijack the preview in unrelated active chats
- [MODEL] FABLE usage policy
- [BUG] Cyber-safeguard block is returned as HTTP 400 invalid_request_error, not a model refusal — endpoint-layer, fires on benign ops across Opus 4.8 / 4.7 / Fable 5 (166 blocks in 11 days, request IDs included)
- [Bug] Excessive refusals on legitimate open-source security implementation tasks
- [Bug] Claude Code bypasses explicit CLAUDE.md push prohibition under autonomous pipeline instruction
- [Bug] Claude Fable 5 classifier false positive on standard refactoring commands with security-unrelated terminology
- [Bug Report] Unable to generate issue title - no bug report content provided
- Feature request: option to disable "Pull request status couldn't be checked" notification for no-PR workflows
- I appreciate you reaching out, but I need clarification. Your message is in Japanese and translates roughly to "That earlier one wasn't really security-related." However, I don't have context about what "that earlier one" refers to, as this appears to be
- [Bug] Model unexpectedly switches to Claude Opus 4 during execution mid-task. Should give option to stop task instead of switching models.
- [BUG] [SECURITY] OAuth session cached after logout — auto-approves login without credential prompt on shared machines
- [BUG] SessionStart hook env files (sessionstart-hook-N.sh) accumulate duplicate lines on re-run — never truncated; can escalate to null-byte Bash tool failure on Windows
- [FEATURE] Agent view: stable sort mode (fixed order + status as indicator) to stop constant reordering
- [BUG] Symlink paths don't match permission patterns, causing repeated prompts
- [BUG] Cowork "VM connection timeout after 60 seconds" — all repair paths exhausted
- [Bug] claude:// URLs do not auto-toggle shell mode like copy-paste commands
- [BUG] Usage limit reached unexpectedly after 429 rate limit errors
- Edit reports success but file unchanged in long sessions (+ batched tool calls dropped, Bash output corruption)
- [FEATURE] Show remote SSH host indicator in Claude Desktop app
- [BUG] the permission prompt gets hidden behind pre-existing text in the input buffer
- [BUG] Claude took liberties it really shouldn't with my global config
- [BUG] Linux: "Couldn't link this session to a computer" — claude-native has no enclave-key implementation (`linux-not-yet-supported`), so td-v2 device registration can never succeed